CSE 373 Data Structures Lecture 6 Stacks and Queues CSE 373 Data Structures Lecture 6

Stacks and Queues - Lecture 6 Readings Reading Sections 3.3 and 3.4 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Stack ADT A list for which Insert and Delete are allowed only at one end of the list (the top) the implementation defines which end is the "top" LIFO – Last in, First out Push: Insert element at top Pop: Remove and return top element (aka TopAndPop) IsEmpty: test for emptyness a tray stack 12/26/03 Stacks and Queues - Lecture 6

An Important Application of Stacks Parsing phase in compilers yields the reverse Polish (postfix) notation: ab+c*d+ (traversal of a binary tree in postorder; see Lecture 7) parse tree d + (a+b)*c+d c * a + b 12/26/03 Stacks and Queues - Lecture 6

Another Important Application of Stacks Call stack in run time systems When a function (method, procedure) is called the work area (local variables, copies of parameters, return location in code) for the new function is pushed on to the stack. When the function returns the stack is popped. So, calling a recursive procedure with a depth of N requires O(N) space. 12/26/03 Stacks and Queues - Lecture 6

Two Basic Implementations of Stacks Linked List Push is InsertFront Pop is DeleteFront (Top is “access” the element at the top of the stack) IsEmpty is test for null Array The k items in the stack are the first k items in the array. 12/26/03 Stacks and Queues - Lecture 6

Linked List Implementation Stack of blobs null node a blob Pointer to next node Pointer to blob 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Array Implementation Stack of blobs A bottom top 1 2 3 4 5 6 7 8 9 10 11 12 4 12 holder = blob pointer array size = number in stack maxsize = max size of stack 12/26/03 Stacks and Queues - Lecture 6

Push and Pop (array impl.) IsEmpty(A : blobstack pointer) : boolean { return A.size = 0 } IsFull(A : blobstack pointer) : boolean { return A.size = A.maxsize; Pop(A : blobstack pointer) : blob pointer { // Precondition: A is not empty // A.size := A.size – 1; return A.holder[A.size + 1]; Push(A : blobstack pointer, p : blob pointer): { // precondition: A is not full// A.size := A.size + 1; A.holder[A.size] := p; 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Linked Lists vs Array Linked list implementation + flexible – size of stack can be anything + constant time per operation - Call to memory allocator can be costly Array Implementation + Memory preallocated + constant time per operation. - Not all allocated memory is used Overflow possible - Resizing can be used but some ops will be more than constant time. 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Insert at one end of List, remove at the other end Queues are “FIFO” – first in, first out Primary operations are Enqueue and Dequeue A queue ensures “fairness” customers waiting on a customer hotline processes waiting to run on the CPU 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Queue ADT Operations: Enqueue - add an entry at the end of the queue (also called “rear” or “tail”) Dequeue - remove the entry from the front of the queue IsEmpty IsFull may be needed 12/26/03 Stacks and Queues - Lecture 6

A Sample of Applications of Queues File servers: Users needing access to their files on a shared file server machine are given access on a FIFO basis Printer Queue: Jobs submitted to a printer are printed in order of arrival Phone calls made to customer service hotlines are usually placed in a queue 12/26/03 Stacks and Queues - Lecture 6

Pointer Implementation Q front rear null Header Not always there front rear 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 List Implementation IsEmpty(Q : blobqueue pointer) : boolean { return Q.front = Q.rear } Dequeue(Q : blobqueue pointer) : blob pointer { // Precondition: Q is not empty // B : blob pointer; B := Q.front.next; Q.front.next := Q.front.next.next; return B; Enqueue(Q : blobqueue pointer, p : blob pointer): { Q.rear.next := new node; Q.rear := Q.rear.next; Q.rear.value := p; 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Array Implementation Circular array front rear Q 0 1 2 3 4 5 6 7 8 9 10 11 4 2 12 holder = blob pointer array size = number in queue front = index of front of queue maxsize = max size of queue rear = (front + size) mod maxsize 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Wrap Around front Q rear 0 1 2 3 4 5 6 7 8 9 10 11 4 10 12 rear = (front + size) mod maxsize = (10 + 4) mod 12 = 14 mod 12 = 2 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Enqueue front Q rear 0 1 2 3 4 5 6 7 8 9 10 11 4 10 12 p 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Enqueue front Q rear 0 1 2 3 4 5 6 7 8 9 10 11 5 10 12 p 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Enqueue Enqueue(Q : blobqueue pointer, p : blob pointer) : { // precondition : queue is not full // Q.holder[(Q.front + Q.size) mod Q.maxsize] := p; Q.size := Q.size + 1; } Constant time! 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Dequeue front Q rear 0 1 2 3 4 5 6 7 8 9 10 11 4 10 12 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Dequeue Q rear front 0 1 2 3 4 5 6 7 8 9 10 11 3 11 12 return 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Try Dequeue Define the circular array implementation of Dequeue 12/26/03 Stacks and Queues - Lecture 6

Stacks and Queues - Lecture 6 Solution to Dequeue Dequeue(Q : blobqueue pointer) : blob pointer { // precondition : queue is not empty // p : blob pointer p := Q.holder[Q.front]; Q.front := (Q.front + 1) mod Q.maxsize; Q.size := Q.size - 1; return p; } 12/26/03 Stacks and Queues - Lecture 6